Removing functions from GenomicRegion that are defined in chromosome_utils. These should probably be redesigned.

Author: andrewdavidsmith

GenomicRegion.cpp

@@ -474,194 +474,3 @@ ReadBEDFile(const string &filename, vector<SimpleGenomicRegion> &the_regions) {
     if (!is_header_line(line) && !is_track_line(line))
       the_regions.push_back(SimpleGenomicRegion(line));
 }
-
-
-static const char *digits = "0987654321";
-static const char *whitespace = " \t";
-
-// define what to do if parts are missing: if no ':' or '-', is whole
-// thing chrom?
-
-void
-parse_region_name(string region_name,
-                  string& chrom, size_t &start, size_t &end) {
-
-  const size_t colon_offset = region_name.find(":");
-
-  // get the chromosome
-  size_t chr_offset = region_name.find_last_of(whitespace, colon_offset);
-  if (chr_offset == string::npos)
-    chr_offset = 0;
-  else
-    chr_offset += 1;
-  chrom = region_name.substr(chr_offset, colon_offset - chr_offset);
-
-  // get the start
-  const size_t start_end = region_name.find("-", colon_offset + 1);
-  const string start_string = region_name.substr(colon_offset + 1,
-                                                 start_end - colon_offset + 1);
-  start = static_cast<size_t>(atoi(start_string.c_str()));
-
-  // get the end
-  const size_t end_end =
-    region_name.find_first_not_of(digits, start_end + 1);
-  const string end_string = region_name.substr(start_end + 1,
-                                               end_end - start_end - 1);
-  end = static_cast<size_t>(atoi(end_string.c_str()));
-}
-
-static size_t adjust_start_pos(const size_t orig_start,
-                               const string &chrom_name) {
-  static const double LINE_WIDTH = 50.0;
-  const size_t name_offset = chrom_name.length() + 2; // For the '>' and the '\n';
-  const size_t preceding_newlines =
-    static_cast<size_t>(std::floor(orig_start / LINE_WIDTH));
-  return orig_start + preceding_newlines + name_offset;
-}
-
-static size_t
-adjust_region_size(const size_t orig_start, const size_t orig_size) {
-  static const double LINE_WIDTH = 50.0;
-  const size_t preceding_newlines_start =
-    static_cast<size_t>(std::floor(orig_start / LINE_WIDTH));
-  const size_t preceding_newlines_end =
-    static_cast<size_t>(std::floor((orig_start + orig_size) / LINE_WIDTH));
-  return (orig_size + (preceding_newlines_end - preceding_newlines_start));
-}
-
-
-void
-extract_regions_chrom_fasta(const string &chrom_name,
-                            const string &filename,
-                            const vector<SimpleGenomicRegion> &regions,
-                            vector<string> &sequences) {
-
-  std::ifstream in(filename.c_str());
-  if (!in)
-    throw runtime_error("cannot read file: " + filename);
-
-  for (auto i = begin(regions); i != end(regions); ++i) {
-
-    const size_t orig_start_pos = i->get_start();
-    const size_t orig_end_pos = i->get_end();
-    const size_t orig_region_size = orig_end_pos - orig_start_pos;
-
-    const size_t start_pos = adjust_start_pos(orig_start_pos, chrom_name);
-    const size_t region_size =
-      adjust_region_size(orig_start_pos, orig_region_size);
-
-    in.seekg(start_pos);
-    char buffer[region_size + 1];
-    buffer[region_size] = '\0';
-    in.read(buffer, region_size);
-
-    std::remove_if(buffer, buffer + region_size,
-                   std::bind2nd(std::equal_to<char>(), '\n'));
-    buffer[orig_region_size] = '\0';
-
-    sequences.push_back(buffer);
-    std::transform(sequences.back().begin(), sequences.back().end(),
-                   sequences.back().begin(), std::ptr_fun(&toupper));
-    assert(i->get_width() == sequences.back().length());
-  }
-}
-
-void
-extract_regions_chrom_fasta(const string &chrom_name,
-                            const string &filename,
-                            const vector<GenomicRegion> &regions,
-                            vector<string> &sequences) {
-
-  std::ifstream in(filename.c_str());
-  if (!in)
-    throw runtime_error("cannot read file: " + filename);
-
-  for (auto i(regions.begin()); i != regions.end(); ++i) {
-
-    const size_t orig_start_pos = i->get_start();
-    const size_t orig_end_pos = i->get_end();
-    const size_t orig_region_size = orig_end_pos - orig_start_pos;
-
-    const size_t start_pos = adjust_start_pos(orig_start_pos, chrom_name);
-    const size_t region_size =
-      adjust_region_size(orig_start_pos, orig_region_size);
-
-    in.seekg(start_pos);
-    char buffer[region_size + 1];
-    buffer[region_size] = '\0';
-    in.read(buffer, region_size);
-
-    std::remove_if(buffer, buffer + region_size,
-                   std::bind2nd(std::equal_to<char>(), '\n'));
-    buffer[orig_region_size] = '\0';
-
-    sequences.push_back(buffer);
-    std::transform(sequences.back().begin(), sequences.back().end(),
-                   sequences.back().begin(), std::ptr_fun(&toupper));
-    if (i->neg_strand())
-      revcomp_inplace(sequences.back());
-    assert(i->get_width() == sequences.back().length());
-  }
-}
-
-void
-extract_regions_fasta(const string &dirname,
-                      const vector<GenomicRegion> &regions_in,
-                      vector<string> &sequences) {
-
-  static const string FASTA_SUFFIX(".fa");
-  assert(check_sorted(regions_in));
-
-  vector<string> filenames;
-  read_dir(dirname, filenames);
-
-  vector<vector<GenomicRegion> > regions;
-  separate_chromosomes(regions_in, regions);
-
-  unordered_map<string, size_t> chrom_regions_map;
-  for (size_t i = 0; i < filenames.size(); ++i)
-    chrom_regions_map[strip_path(filenames[i])] = i;
-
-  for (size_t i = 0; i < regions.size(); ++i) {
-    // GET THE RIGHT FILE
-    const string chrom_name(regions[i].front().get_chrom());
-    const string chrom_file(chrom_name + FASTA_SUFFIX);
-    unordered_map<string, size_t>::const_iterator f_idx =
-      chrom_regions_map.find(chrom_file);
-    if (f_idx == chrom_regions_map.end())
-      throw SMITHLABException("chrom not found:\t" + chrom_file);
-    extract_regions_chrom_fasta(chrom_name, filenames[f_idx->second],
-                                regions[i], sequences);
-  }
-}
-
-void
-extract_regions_fasta(const string &dirname,
-                      const vector<SimpleGenomicRegion> &regions_in,
-                      vector<string> &sequences) {
-
-  static const string FASTA_SUFFIX(".fa");
-  assert(check_sorted(regions_in));
-
-  vector<string> filenames;
-  read_dir(dirname, filenames);
-
-  vector<vector<SimpleGenomicRegion> > regions;
-  separate_chromosomes(regions_in, regions);
-
-  unordered_map<string, size_t> chrom_regions_map;
-  for (size_t i = 0; i < filenames.size(); ++i)
-    chrom_regions_map[strip_path(filenames[i])] = i;
-
-  for (size_t i = 0; i < regions.size(); ++i) {
-    // GET THE RIGHT FILE
-    const string chrom_name(regions[i].front().get_chrom());
-    const string chrom_file(chrom_name + FASTA_SUFFIX);
-    unordered_map<string, size_t>::const_iterator f_idx =
-      chrom_regions_map.find(chrom_file);
-    if (f_idx == chrom_regions_map.end())
-      throw SMITHLABException("chrom not found:\t" + chrom_file);
-    extract_regions_chrom_fasta(chrom_name, filenames[f_idx->second],
-                                regions[i], sequences);
-  }
-}

GenomicRegion.hpp

@@ -419,11 +419,6 @@ WriteBEDFile(const std::string filename,
   out.close();
 }
 
-
-void
-parse_region_name(std::string region_name,
-                  std::string &chrom, size_t &start, size_t &end);
-
 template <class T>
 std::string
 assemble_region_name(const T &region) {
@@ -438,28 +433,5 @@ assemble_region_name(const T &region, const std::string sep) {
           smithlab::toa(region.get_end()));
 }
 
-void
-extract_regions_chrom_fasta(const std::string &chrom_name,
-                            const std::string &filename,
-                            const std::vector<GenomicRegion> &regions,
-                            std::vector<std::string> &sequences);
-
-
-void
-extract_regions_chrom_fasta(const std::string &chrom_name,
-                            const std::string &filename,
-                            const std::vector<SimpleGenomicRegion> &regions,
-                            std::vector<std::string> &sequences);
-
-void
-extract_regions_fasta(const std::string &dirname,
-                      const std::vector<SimpleGenomicRegion> &regions_in,
-                      std::vector<std::string> &sequences);
-
-void
-extract_regions_fasta(const std::string &dirname,
-                      const std::vector<GenomicRegion> &regions_in,
-                      std::vector<std::string> &sequences);
-
 
 #endif